DataStage job stuck in a running state

[orondedata]

I have been support DataStage for 2 weeks now and I have had a couple jobs get stuck in a running state. What is the best automated way to monitor job that never complete since they don't send a success or failure email.

[chulett]

It's important to understand that the status of a job is basically just a 'flag', something that gets set when the job starts and adjusted when the job finishes. So server crashes or unusual situations where a job may just core dump and fall over dead can leave it stuck forever in a 'Running' status.

First thing to do is check to see if it is actually running or not. From the UNIX command line issue a "ps -ef | grep phantom" and see if you can see any processes associated with the job name in question. If nothing shows up, then you can use the Director client to reset that status flag. Go to the Job menu and choose Clear Status File. It will ask twice if you're really sure you want to do this as it is not recommended to do this to a job that is actually running, but after you say 'yes' it will set the job's status back to Compiled and you can go from there.

[austin_316]

one of my job is getting stuck and i have given the command "ps -ef | grep phantom" in the unix. i am able to see a process assocaited with my job. how can i stop that job! please help me in this.

[Sainath.Srinivasan]

Did you try the stop button in director ?

Are you sure that the process is running but not doing anything ?

Is it in any deadlock situation or doing an after-job script or waiting for something ?

I will recommend checking the process action before killing or stopping it.

[ray.wurlod]

Check for lock contention in any database to which the job connects. A job will wait up to forever for a database lock (or table or row lock) to be released.

[Sreenivasulu]

Ray is correct. This happened for me last week and the issue was that the 'database was not in right shape'. May be for you it might be some other issue.
But first
1.you need to stop the long-running job
2.If it has run earlier properly then run-again
3.If step 2 fails check the enviroment

Regards
Sreeni

[austin_316]

Yes as suggested by Ray the issue was with some lock on a table. i have modified the rows of the table manually and forgot to commit the db. there was a before job script which is trying to update hte same row in the same table. but since the lock was already made on the row the job was waiting for it to be released. when i commited the db then the process continued.
This solution i got for one situation.but again when again the job was in runnning condition for a long time i tried the same process. but still the problem persisted. so may there was some other reason for this. Not able to find it though [/code]

[austin_316]

Yes as suggested by Ray the issue was with some lock on a table. i have modified the rows of the table manually and forgot to commit the db. there was a before job script which is trying to update hte same row in the same table. but since the lock was already made on the row the job was waiting for it to be released. when i commited the db then the process continued.
This solution i got for one situation.but again when again the job was in runnning condition for a long time i tried the same process. but still the problem persisted. so may there was some other reason for this. Not able to find it though

[ray.wurlod]

Have your DBA check for deadlocks and lock waits in the database.